Title: On predicting machined part accuracy from CNC machine errors using artificial neural networks

Authors: Marios-Christos Koutsogiannis; George-Christopher Vosniakos

Addresses: School of Mechanical Engineering, Manufacturing Technology Laboratory, National Technical University of Athens, Heroon Politehniou 9, 15773, Athens, Greece ' School of Mechanical Engineering, Manufacturing Technology Laboratory, National Technical University of Athens, Heroon Politehniou 9, 15773, Athens, Greece

Abstract: Geometric errors of computer numerical control (CNC) machines have a direct effect on the finished product accuracy. This paper proposes a method of correlation between position deviations of the cutting tool path in a 3-axis machine and the accuracy of part features, by example of concentricity and circularity of nominally cylindrical surfaces on a benchmark ISO test piece. The true position of the cutting tool is derived from a kinematic chain model incorporating all 21 geometric errors of the machine, fully mapped using laser doppler metrology. 16 machining tests were executed at different positions on the machine workspace. At every position circularity and concentricity of the considered features were calculated according to the kinematic model and also measured on a coordinate measuring machine (CMM). Calculated and measured accuracy values were used to train artificial neural networks as accuracy predictors.

Keywords: CNC; computer numerical control; geometric errors; metrology; CMM; coordinate measuring machine; CNC kinematic chains; ANNs; artificial neural networks; machining; machined test piece; laser metrology; CNC error mapping.

DOI: 10.1504/IJMMS.2023.133394

International Journal of Mechatronics and Manufacturing Systems, 2023 Vol.16 No.2/3, pp.280 - 300

Received: 15 Feb 2023
Accepted: 01 Jun 2023
Published online: 14 Sep 2023 *